Top US AI labs urge Washington to lead frontier regulation
The leaders of OpenAI, Google DeepMind and Anthropic have united to call for a single US-led regulator for advanced AI, a push that sidelines European standard-setting and risks entrenching American tech dominance.
The chief executives of OpenAI, Google DeepMind and Anthropic have separately published proposals over the last five weeks. They are united in urging the creation of a single US-led body to regulate frontier AI models.
Despite fierce market competition, the three rivals share a near-identical diagnosis. They want independent outsiders to test the most capable models before release, replacing the industry's old habit of self-reporting. Crucially, all three want Washington to take the lead, explicitly rejecting a patchwork of state or rival national rules.
For European policymakers, this alignment is a direct challenge to Brussels' ambitions. By insisting that the US must set global standards, these American labs are positioning Washington as the primary arbiter of AI safety. This effectively marginalises the EU’s own regulatory framework just as it begins to take effect.
The rivals diverge on the exact mechanism. Anthropic’s Dario Amodei wants a federal agency akin to the FAA that can block releases outright. DeepMind’s Demis Hassabis prefers a FINRA-style, industry-funded but government-overseen body starting with voluntary reviews. Writing in the Financial Times, OpenAI’s Sam Altman argued for an IAEA-style international forum using market access as leverage.
Hassabis’s plan drew rare cross-industry praise. Altman called it “thoughtful.” Microsoft’s Satya Nadella said the goal was “a frontier ecosystem that promotes innovation and choice.” Elon Musk, despite recent spats with Altman, deemed it “a good starting point,” while Anthropic’s Jack Clark called the framework “excellent.”
The timing mirrors recent US government action. Over the same five weeks, Washington restricted or delayed Anthropic’s Fable and Mythos models, followed by OpenAI’s GPT-5.6, over cyber fears. While the Trump administration remains publicly anti-regulation, it is privately less sure a hands-off approach can hold. Meta’s Mark Zuckerberg is reportedly drafting his own memo.
The proposed safety net comes with a significant market distortion risk. Critics warn of regulatory capture. OpenAI, Google and Anthropic already possess the legal teams, security infrastructure and Washington lobbying networks to navigate a heavy certification process. Startups and open-source developers do not. The same executives who sparred at the G7 weeks ago are now lobbying for rules that could quietly entrench the biggest labs.
